I have a linux server with 2 nic's both in a separate subnet (172.21.3.x
the 192.168.10.x is connected to another server via a crossed utp cable.
the 172.21.3 is our local lan.
if i ping the 192.168.10.x interface on the server from my local
workstation (which is in the 172.21.3.x range), i get an immediate reply.
How does this happen ? does the linux server hear the request for
192.168.10.x on his 172.21.3.x interface and thinks, "hey, that's my other
interface, i'll reply" ?
in the routing table, there is no reference to the 192.168.10.x subnet,
and our default gateway also doesn't know about this subnet.
if my assumptions are correct, is it possible to stop the linux server
from responding to icmp queries not directed to its correct (same subnet)
I don't use iptables (yet). i noticed the same behaviour with the
/proc/sys/net/ipv4/ip_forward parameter either set to 0 or 1.